OpenAI Hires Former Army Secretary Dan Driscoll to Lead a New National Security Push

OpenAI has hired Dan Driscoll, the 26th secretary of the US Army, as its senior counselor for national security modernization, a position the company created for him and announced on Monday. The role puts a recently departed Pentagon civilian leader inside the lab that is simultaneously trying to sell AI to governments, defend its own models from attack and argue that it can be trusted with sensitive work.

Driscoll ran the Army for about 18 months, overseeing close to one million active-duty, National Guard and Reserve soldiers along with more than 265,000 civilian employees. His tenure focused on pulling new technology into the force faster, rewriting procurement rules and pushing modernization programs that had stalled. He had earlier served as acting director of the Bureau of Alcohol, Tobacco, Firearms and Explosives, and before that was an Army officer and a lawyer.

He resigned in late August. The White House confirmed the departure at the time without giving a reason, and reporting since has described friction between Driscoll and Defense Secretary Pete Hegseth, a characterization neither man has detailed publicly.

At OpenAI, Driscoll will advise senior leadership and work with government officials, industry, academia and allied institutions on how advanced AI can strengthen national security and cybersecurity. The company said he will look for the intersection of operational requirements, institutional change and emerging model capabilities, and will help internal teams understand how national security agencies actually operate rather than how vendors imagine they do.

"National security requirements are changing faster than at any point in the past 50 years," Driscoll said in a statement. He argued that AI can help frontline staff cut the paperwork burden, integrate large volumes of information and get the right answer to the right person faster. Chief executive Sam Altman framed the hire around what OpenAI calls democratic AI, saying the United States has an opportunity to work with other countries to spread it.

The appointment extends Daybreak, the company's program for working with allies on cyber resilience, critical infrastructure protection and public service security. It also lands in the middle of a broader repositioning in which OpenAI has been hardening its own defenses while arguing that governments, not just consumers, should be its next large customer base.

What it means: frontier labs are increasingly hiring the people who write the requirements, not only the people who write the code. Driscoll's job is part sales and part translation, and the harder half will be convincing allied governments that a company which has told investors it may imminently cross a recursive self-improvement threshold can also be a dependable supplier of defense-grade software.